VaaniX
Catch AI voice-clone scams during the call — before the money moves.
Generative AI can replicate any family member’s voice from just 3 seconds of audio. VaaniX listens to seconds 5 to 15 of suspicious calls, verifies biometric vocal authenticity using deep acoustic models, and halts fraudulent transfers in real time.
+91 98765 43210
Unknown Caller · 00:14AI VOICE DETECTED
Acoustic biometric anomaly detected. High synthetic clone probability.
Instant Telegram alert sent to your registered account
The AI Voice-Clone Threat in India
Criminal syndicates weaponize generative voice cloning to impersonate sons, daughters, and officials. Traditional spam call filters fail completely because the voice sounds authentic to the parent.
of Indian adults have faced, or know someone hit by, an AI voice scam (global average: 25%)
of Indian victims lost money; 48% of them lost over ₹50,000 to fraud
of audio was enough for an 85% voice match in security evaluations
Indians lost ₹22,845 crore to cyber fraud in 2024, up from ₹7,465 crore in 2023
of deepfake speech samples were missed by people in a 529-person study, even after training
How the Voice-Clone Scam Plays Out
Voice Harvesting
Scammers capture 3 seconds of voice from social media videos or a 2-second blank robocall.
Neural Cloning
Deep generative AI clones the family member’s pitch, tone, vocal cadence, and accent in seconds.
The Distress Call
“Mummy, accident ho gaya, police station mein hoon, jaldi paise bhejo!”
Panic & Payment
The terrified parent sends funds via UPI before hanging up or verifying with family.
10 Seconds to Absolute Certainty
VaaniX does not tap or record entire phone calls. It monitors a single 10-second window (seconds 5 to 15), processes acoustics and transcript in parallel, and scores risk transparently.
Call Connects
0s · Ringing connects
Skip Greetings
0s–5s · Ignore intro
10s Recording
5s–15s · Mono 16kHz
AI Detected
Risk ≥ 60 & Voice ≥ 50
Real Voice
Risk < 40
Under The Hood (Biometric & Semantic Dual Verification)
VaaniX records a 10-second clip of the call and checks it two ways: the Nes2Net-X model tells whether the voice is real or AI-generated, and the transcript shows who the caller claims to be and picks up any scam words. Both results are combined into one risk score from 0 to 100, and a high score instantly triggers the alerts.
Three Instant Lines of Defense
When a voice clone is identified, alerts fire simultaneously across the screen, your Telegram, and to the real family member being impersonated.
AI VOICE DETECTED
Do NOT transfer money or share OTPs.
On-Screen Intervention
Direct floating overlay informs the victim of synthetic speech during the active call.
Telegram Alert to You
Instant notification with pre-drafted reporting links and 1-tap helpline access.
Impersonation Shield
The real family member receives a prompt: “Is it you?” Tapping “No” immediately confirms the scam.
Family Circle & Live Transcript Inspection
Add your children, spouse, and parents. When a caller claims to be your son, VaaniX cross-references your family relationships and verifies the claim in real time.
Protected Family Circle
Each member joins via a simple Telegram invite link (t.me/Prevent_Scambot).
Call Transcript
Mummy, mera car accident ho gaya hai.
Police khadi hai, jaldi se 25,000 paise upi bhejo.
Papa ko mat batana abhi, emergency hai.
Natural language extraction detects family terminology (“Mummy” → caller claims to be child; “Beta” → claims to be parent) and fuzzy matches names against your registered group members with Levenshtein distance ≤ 1.
Scam Radar & Payment Guard
Collective protection across callers before pickup, and automated financial friction after the call.
Scam Radar Community Lookup
Fast (< 300ms)While the phone is ringing, the caller’s number is queried against the decentralized Scam Radar index. If 3 or more users flagged the number in 24 hours, a warning pop-up alerts you before answering.
Most reported scam category: Fake Police / Arrest Impersonation
Note: Numbers can be spoofed. VaaniX warns, it never blocks. No reporter identities are ever exposed.
Payment Guard Intercept
Coming in Android AppFor 30 minutes following an AI-flagged call, opening any UPI or banking application (Google Pay, PhonePe, Paytm, BHIM, YONO) initiates a mandatory 60-second cooldown screen.
“You just finished an AI voice-flagged call. Pause and call your family member on a trusted number before transferring funds.”
One-Tap Reporting & Developer API
Streamline reporting to law enforcement and financial institutions, or embed deepfake audio detection directly into fintech mobile applications.
Official Escalation Channels
Pre-fills official complaint email with cryptographic audio SHA-256 hash
Drafts emergency transaction freeze notice to your registered bank fraud desk
One-tap dialer for the National Cyber Crime Helpline (stops financial flight)
Provides user feedback to refine scoring and exclude numbers from radar
Developer REST API
Read API Docs/api/v1/analyzeSubmit 10s audio WAV; returns biometrics, transcript, and risk score
/api/v1/complaintGenerates structured JSON complaint package with signed evidence link
/api/v1/radar/lookupSub-300ms number reputation query for telecom and caller-ID apps
satyavoice-sdkEmbeddable Android module for banking and UPI applications (Planned)
Who VaaniX Protects
Engineered for Indian families, financial institutions, and cyber crime investigation units.
Senior Citizens & Families
Shields parents and grandparents from extortion calls faking their children's voices in mock accidents, hospital emergencies, or digital arrest scams.
Banks & UPI Applications
Integrate acoustic scoring and automated transaction cooldowns before victims authorize irrevocable immediate payments.
Telecoms & Law Enforcement
Generates immutable SHA-256 evidence digests and structured complaint formats for state cyber crime cells and helpline 1930 operators.
Your voice is your identity.
VaaniX protects it.
Real-time acoustic AI voice-clone verification and family impersonation shield.